Atos at Nvidia GTC
Atos, Diamond sponsor, invites you to follow digital sessions from April 12 – 16. This event provides a unique platform for AI innovators, technologists and creatives.
To register, click here or fill in the form at the bottom of the page.
Advancing Exascale – Faster, Smarter & Greener!
Session ID SS32805 – Live chat Q&A session
Wednesday 14 April Broadcast at 1:00 – 1:40pm CET
CTO, HPC & Quantum, Atos
Exascale systems are orders of magnitude more powerful than currently deployed HPC systems. The performance increase relies mainly on the use of computing accelerators such as GPUs. These accelerators also enable new AI algorithms used in complex workflows to improve data assimilation, data analysis, or computing itself and optimize the HPC datacenter resources utilization. On the energy consumption side, decarbonization of the Exascale systems is the challenge for the next decade.
Atos High Performance AI
and NVIDIA DGX SuperPOD
Session ID SS32969 – On-demand session
Atos AI Distinguished Expert, HPC-AI Sales Operations Lead, Atos
Though AI has been widely demystified compared with its debut, its full materialization in real field implementation entails diversified challenges. C. Bourrasset will reveal how Atos pioneer to successfully architect, build and deliver large scale AI Infrastructure based on NVIDIA DGX SuperPOD reference design. Through live customer cases and lesson learnt from the field, Cedric will detail Atos added value on top of DGX SuperPOD Solution such as Atos Data Science Platform (CODEX AI SUITE) to leverage SuperPOD performance at scale, predictive maintenance software to enhance system uptime, and will show how to build & deliver a SuperPOD in the middle of the COVID Pandemic.
Real-Time Video Analytics: Why is Edge computing decisive for Public Transports?
Session ID SS32971 – On-demand session
Head of Product, Edge & AI, Atos
AI Product Manager, Atos
Client executive, Public Sector, Atos
This session aims to present how the usage of edge computing is imperative to leverage the potential of computer vision for transports and how GPUs are key. Also come to discover how to orchestrate and seamlessly deploy secure end-to-end Vision solutions by using Nvidia Fleet Command Manager. BullSequana Edge server is fully optimized to support Fleet Commands secure one touch deployment and encryption options. We will share best practices in factory pre-load to secure processes, reduce costs & manage edge services.
AI Industrialization in an R&D lab: challenges, solutions & feedback
Session ID SS32970 – On-demand session
Head of AI, Distinguished Expert, Member of Atos Scientific Community, Atos
AI Chief Architect, Senior Expert, Atos
This session aims to share feedback from the ATOS software R&D lab on the requirements for common tools and methods to improve the quality and efficiency of the internal AI developments and more broadly their industrialization. The session will present the different challenges and implemented solutions: – Shared training platform & infrastructure: using Atos Codex AI Suite FastML Engine, NVIDIA GPU cluster and Docker – AI reproducibility & traceability: including internal reproducibility checklist, automated tracking using MLFlow, dataset versioning, automated trainings with Jenkins This session is intended to architects and AI developers.